High Performance Mysql - Friday Tech Talks at Squareboat
0 likes218 views
The document discusses various strategies for optimizing queries and indexes in MySQL for high performance. It covers normalizing schemas, using different types of indexes like B-tree and hash indexes, indexing strategies like isolating columns and using prefix/suffix indexes. It also discusses covering indexes, cluster indexes, optimizing ALTER TABLE queries, analyzing slow queries, restructuring queries, and basics of query execution and optimization in MySQL.
![MySQL join execution strategy
●
mysql> SELECT tbl1.col1, tbl2.col2
FROM tbl1 INNER JOIN tbl2
USING(col3) where tbl1.col1 IN (5,6)
outer_iter = iterator over tbl1 where col1 in (5,6)
outer _row = outer_iter.next
while outer_row
inner_iter = iterator over tbl2 where col3 =outer_row.col3
inner_row = inner_iter.next
while inner_row
output [outer_row.col1,inner_row.col2 ]
inner_row = inner_iter.next
end
outer_row = inner_iter.next
end](https://image.slidesharecdn.com/techtalkhighperformancemysql-190923161932/85/High-Performance-Mysql-Friday-Tech-Talks-at-Squareboat-17-320.jpg)
High Performance Mysql - Friday Tech Talks at Squareboat
- 2. ● Normalization of Schema. ● Indexes of Schema. Schema optimization
- 3. B-Tree Indexes CREATE TABLE PEOPLE ( Last_name varchar(50) NOT null first_name varchar(50) NOT null dob varchar(50) NOT null gender enum (50) NOT null Key( last_name,first_name,dob) Allen Cuba 1960-01-0 1 Astaire Angelia 1980-03-04 Barrymore julia 2000-05-16 akroyd christan 1958-12-07 akroyd Debbie 1990-03-18 Allen Cuba 1960-01-01 Allen Kim 1930-07-12 Barrymore Julia 2000-05-16 Basinger Viven 1976-12-08 Type of Indexes
- 4. Types of Queries that can use a B-Tree Index ● ● ● ● ● ● Limitations ● ● Indexed key(‘last_name’, ’first_name’, ’dob’)
- 5. Hash Indexes ● ● ● fname lname Baron Lentz Peter Schwartz Vadim Zaitsev Arjen Tkachenko Ex: f(Arjen) = 2458 Slot lname 2323 Pointer to row 1 2458 Pointer to row 4 2537 Pointer to row 2 8784 Pointer to row 3
- 6. Indexing strategies for high performance ❏ Isolate the Column ● “Isolating” the column means, indexed column should not be a part of an expression or be inside a function in the query. ● Example Incorrect Correct Select * from actor where action_id+1 =5 Select * from actor where action_id = 5 Select . . . where TO_DAYS( CURRENT_DATE) - TO_DAYS( date_col) <= 10 Select . . . where date_col >= DATE_SUB(CURRENT_DATE,INTERVAL 10 DAY )
- 7. ❏ Prefix & Suffix Index and Index selectivity ● For a column of very long length string we can use prefix of indexing. ● But what would be the length of prefix to maintain its good selectivity ? Steps to calculate selectivity : 1. Find full column selectivity. Ex : Select COUNT( DISTINCT(City)) / Count(*) from city_table (0.0312). 2. Now calculate selectivity of each length Select COUNT( DISTINCT LEFT(City,3)) / Count(*) from city_table (0.0329). Select COUNT( DISTINCT LEFT(City,4)) / Count(*) from city_table (0.0293). Select COUNT( DISTINCT LEFT(City,7)) / Count(*) from city_table (0.0310). City Count London 65 Hiroshima 49 Mumbai 48 Delhi 48 Shanghai 48 Kolkata 47 Chennai 47 New York 45 7 length selectivity value is close to full column selectivity. mysql> ALTER TABLE city Add key (city(7))
- 8. ❏ Cluster Indexes ● Cluster indexing is special approach to data storage. ● Index row itself stored in the index leaf page. ● One cluster index per table. 11 21 91 Akroyd Christian 1958-12-07 1 Akroyd Debbie 1990-03-18 2 Allen Cuba 1960-01-01 11 Allen Kim 1930-07-12 12 Barrymore Julia 2000-05-16 92 Basinger Viven 1976-12-08 93 Advantages ● ● Disadvantages ●
- 9. Comparison of InnoDB and MyISAM data layout MyISAM InnoDB It store the rows on disk in order they were inserted It store the rows on disk in cluster organization. Primary index in leaf node store pointer to row Primary index in leaf node store row itself. Secondary index in leaf node store pointer to row. Secondary index in leaf node store primary key of row.
- 11. ❏ Covering Indexes ● Mysql use this index to retrieve a column data instead of row data. ● It is mostly used in InnoDB data layout ● It can dramatically improve the performance. Example : let a comments table indexed with (post_id) Without using covering Index With using covering index Select * from comments where post_id =’1’ limit 10000,10 Select * from comments (Select id from comments where post_id =’1’ limit 10000,10 ) t1 ,comments t2 where t1.id= t2.id
- 12. Speeding Up ALter Table How Alter table actually perform ? ● ● Modifying the .frm file
- 13. Query Optimisation and Performance Retrieving more rows than you need Accessing more column than you need. Analyze a poorly performing query Query cost metrics : ● ●
- 14. Ways To Restructure Queries ❏ Complex vs simple queries ● ● ❏ Chopping Up a query ● ● ❏ Joins decomposition > SELECT * FROM tag JOIN tag_post ON tag_post.tag_id = tag.id JOIN post ON tag_post.post_id = post.id Where tag.tag =’mysql’ > SELECT * FROM tag WHERE tag =’’mysql’ > SELECT * FROM tag_post WHERE tag_id = 1234 > SELECT * FROM post WHERE post_id IN (123,456).
- 16. Query Optimization Process ❏ Parser and the preprocessor ● ● ❏ Query Optimizer ● ● ● ●
- 17. MySQL join execution strategy ● mysql> SELECT tbl1.col1, tbl2.col2 FROM tbl1 INNER JOIN tbl2 USING(col3) where tbl1.col1 IN (5,6) outer_iter = iterator over tbl1 where col1 in (5,6) outer _row = outer_iter.next while outer_row inner_iter = iterator over tbl2 where col3 =outer_row.col3 inner_row = inner_iter.next while inner_row output [outer_row.col1,inner_row.col2 ] inner_row = inner_iter.next end outer_row = inner_iter.next end
- 18. ❏ Join Optimizer Execution 1 Execution 2 TABLE actor :200 rows TABLE film : 951 rows TABLE film_actor :1 row TABLE film_actor :1 row TABLE film :1 row TABLE actor :1 row actor film_actor film film film_actor actor
- 19. Limitation of MySQL Query Optimizer ❏ Correlated Subqueries : How we think MySQl execute it: But what optimizer actually do film_actor.film_id = film_actor.film_id
- 20. ❏ WildCards ● ● ❏ Push predicates into the OUTER JOIN ● ●
- 21. ❏ Duplicate Constant Condition in JOIN ● ● ❏ Avoid sorting on 2 column of different table in JOIN
- 22. THANK YOU